Many data analysts are seeking to use the large amounts of sensor data generated by the Internet of Things (IoT). Unfortunately most of the measurements taken by these sensors are imprecise. Whether analysts are trying to predict machine failure or the location of an earthquake, knowledge of this imprecision is key to obtaining reliable results.
GeoIntelligence analysts use aerial imagery and geospatial data to classify objects within a landscape. This data is imprecise - GPS accuracy is very good but imperfect. Even small errors in the input can magnify when run through a deep learning algorithm such as those used by Intelligence Agencies. Dubito could easily help these analysts to account for uncertainty in their classifications.
Whether the analyst is trying to predict emergency room wait times or the likelihood of success of a particular treatment, a good portion of the data being fed to the algorithm is imprecise. This is one of the most important applications of Dubtio because of the strong need for proper risk assessment when dealing with healthcare situations.
In addition to the use cases listed above, Dubito can be applied to more general types of problems traditionally solved under the assumption of little to no uncertainty. The following presentation explores these concepts.